The aim of this study was to evaluate the effect of exercise on pregnancy outcome in streptozotocin-induced diabetic Wistar rats (n = 11 animals/group). These animals were randomly assigned to sedentary (G1) and exercised groups, beginning from day 0 (G2) or 7 (G3) to day 20 of pregnancy. The moderate exercise was a swimming programme. At day 21 of pregnancy, all rats were anaesthetized and killed to obtain pregnancy outcome data. All rats presented glycaemia higher than 300 mg/dl, regardless of the exercise training. The G3 group showed higher live fetus number per implantation site and lower resorption number per implantation site compared with the G1 group. The fetal and placental mean weights per litter and the total number of ossification sites were significantly lower in the exercised groups (P < 0.05). Placental index was lower in the G2 and G3 groups compared with the G1 group. The occurrence of skeletal anomalies indicated that exercise increased the number of altered fetuses. Thus, moderate exercise achieved better outcomes by increasing the number of live births and decreasing resorption. However, exercise increased skeletal anomalies and decreased fetal and placental weights.